A. Hanss, E. Liu, Muhammad Rizwan Abdullah, G. Elger
{"title":"基于瞬态热分析和校正有限元模型的LED封装失效识别","authors":"A. Hanss, E. Liu, Muhammad Rizwan Abdullah, G. Elger","doi":"10.1109/EUROSIME.2019.8724558","DOIUrl":null,"url":null,"abstract":"Transient thermal analysis (TTA) by experimental thermal impedance $(Z_{th})$ measurements and data simulation by transient finite element (FE) simulation are suited to investigate the thermal path and mechanical integrity of electronic devices. After calibration, the FE model can be used for failure mode analysis. In this paper first a FE model for blue Flip Chip (FC) LEDs is set-up and calibrated to the experimental data. To reduce the ambiguity of the parameter identification within the calibration process caused by the relative large number of model parameters in a transient FE simulation several sets of experimental data are measured and simulated: the LED assembled on different substrates, i.e. an AlN ceramic and a standard Al-IMS board, and similar LEDs, i.e. same LED type but different sized light emitting area (1 mm2 and 0.5 mm2). The model is calibrated in the time domain using b(z), i.e. the logarithmic time derived of $Z_{th}(z= {ln(t))}$. The least square residuum of simulated and experimental data is minimized. Different approaches of fitting are compared, i.e. fitting $b(z), Z_{th}(t)$ and the structure function, resulting in b(z) being a good option for the LED under investigation. The calibration is obtained by an optimizer in an automated workflow, i.e. using ANSYS coupled with optiSLang.After calibration the model is used to identify failures during accelerated stress test. An automatic TTA tester is used to detect changes of the Zth(t) during accelerated lifetime testing, e.g. temperature shock testing.By implementing failure into the calibrated model, e.g. cracks in the solder joint, delamination of the EPI from the redistribution layer and degradation of the dielectric layer of the Al-IMS, the calibrated model is fitted to the b(z) curves of the aged samples. The strength of the approach is demonstrated by extracting the quantitative changes of the physical parameters from the fitted model in an automatic way.","PeriodicalId":357224,"journal":{"name":"2019 20th International Conference on Thermal, Mechanical and Multi-Physics Simulation and Experiments in Microelectronics and Microsystems (EuroSimE)","volume":"59 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2019-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"6","resultStr":"{\"title\":\"Failure Identification in LED packages by Transient Thermal Analysis and Calibrated FE Models\",\"authors\":\"A. Hanss, E. Liu, Muhammad Rizwan Abdullah, G. Elger\",\"doi\":\"10.1109/EUROSIME.2019.8724558\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Transient thermal analysis (TTA) by experimental thermal impedance $(Z_{th})$ measurements and data simulation by transient finite element (FE) simulation are suited to investigate the thermal path and mechanical integrity of electronic devices. After calibration, the FE model can be used for failure mode analysis. In this paper first a FE model for blue Flip Chip (FC) LEDs is set-up and calibrated to the experimental data. To reduce the ambiguity of the parameter identification within the calibration process caused by the relative large number of model parameters in a transient FE simulation several sets of experimental data are measured and simulated: the LED assembled on different substrates, i.e. an AlN ceramic and a standard Al-IMS board, and similar LEDs, i.e. same LED type but different sized light emitting area (1 mm2 and 0.5 mm2). The model is calibrated in the time domain using b(z), i.e. the logarithmic time derived of $Z_{th}(z= {ln(t))}$. The least square residuum of simulated and experimental data is minimized. Different approaches of fitting are compared, i.e. fitting $b(z), Z_{th}(t)$ and the structure function, resulting in b(z) being a good option for the LED under investigation. The calibration is obtained by an optimizer in an automated workflow, i.e. using ANSYS coupled with optiSLang.After calibration the model is used to identify failures during accelerated stress test. An automatic TTA tester is used to detect changes of the Zth(t) during accelerated lifetime testing, e.g. temperature shock testing.By implementing failure into the calibrated model, e.g. cracks in the solder joint, delamination of the EPI from the redistribution layer and degradation of the dielectric layer of the Al-IMS, the calibrated model is fitted to the b(z) curves of the aged samples. The strength of the approach is demonstrated by extracting the quantitative changes of the physical parameters from the fitted model in an automatic way.\",\"PeriodicalId\":357224,\"journal\":{\"name\":\"2019 20th International Conference on Thermal, Mechanical and Multi-Physics Simulation and Experiments in Microelectronics and Microsystems (EuroSimE)\",\"volume\":\"59 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2019-03-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"6\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"2019 20th International Conference on Thermal, Mechanical and Multi-Physics Simulation and Experiments in Microelectronics and Microsystems (EuroSimE)\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1109/EUROSIME.2019.8724558\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"2019 20th International Conference on Thermal, Mechanical and Multi-Physics Simulation and Experiments in Microelectronics and Microsystems (EuroSimE)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/EUROSIME.2019.8724558","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
Failure Identification in LED packages by Transient Thermal Analysis and Calibrated FE Models
Transient thermal analysis (TTA) by experimental thermal impedance $(Z_{th})$ measurements and data simulation by transient finite element (FE) simulation are suited to investigate the thermal path and mechanical integrity of electronic devices. After calibration, the FE model can be used for failure mode analysis. In this paper first a FE model for blue Flip Chip (FC) LEDs is set-up and calibrated to the experimental data. To reduce the ambiguity of the parameter identification within the calibration process caused by the relative large number of model parameters in a transient FE simulation several sets of experimental data are measured and simulated: the LED assembled on different substrates, i.e. an AlN ceramic and a standard Al-IMS board, and similar LEDs, i.e. same LED type but different sized light emitting area (1 mm2 and 0.5 mm2). The model is calibrated in the time domain using b(z), i.e. the logarithmic time derived of $Z_{th}(z= {ln(t))}$. The least square residuum of simulated and experimental data is minimized. Different approaches of fitting are compared, i.e. fitting $b(z), Z_{th}(t)$ and the structure function, resulting in b(z) being a good option for the LED under investigation. The calibration is obtained by an optimizer in an automated workflow, i.e. using ANSYS coupled with optiSLang.After calibration the model is used to identify failures during accelerated stress test. An automatic TTA tester is used to detect changes of the Zth(t) during accelerated lifetime testing, e.g. temperature shock testing.By implementing failure into the calibrated model, e.g. cracks in the solder joint, delamination of the EPI from the redistribution layer and degradation of the dielectric layer of the Al-IMS, the calibrated model is fitted to the b(z) curves of the aged samples. The strength of the approach is demonstrated by extracting the quantitative changes of the physical parameters from the fitted model in an automatic way.